Created by [gnoownow10](https://github.com/gnoownow10). docx, doc, hwp to html.

anything2html.php

<?php
/**
 * @see dependencies: libreoffice, pyhwp https://pythonhosted.org/pyhwp/ko/
 * Class Anything2html
 */
class Anything2html {
    static $hwp5html;
    static $libreoffice;
    static $error;

    static function convert($content, $extension, $default_handler, $error_handler) {
        static::$error = false;
        $handler = "{$extension}_handler";
        $can_handle = method_exists(__CLASS__, $handler);
        if ($can_handle) {
            try {
                return call_user_func([__CLASS__, $handler], $content);
            } catch (ExecFailedException $e) {
                static::$error = true;
                return call_user_func($error_handler, $e);
            }
        } else {
            static::$error = true;
            return call_user_func($default_handler, $content);
        }
    }

    static function executable_exists($cmd) {
        exec("which $cmd", $output, $exit_code);
        return (bool) $output;
    }

    static function get_body_inner_html($html) {
        $dom = new DOMDocument();
        $dom->loadXML($html);
        $body = $dom->getElementsByTagName('body')->item(0);
        $inner_html = '';
        foreach ($body->childNodes as $child) { $inner_html .= $dom->saveHTML($child); }
        return $inner_html;
    }

    static function txt_handler($content) {
        $content = htmlentities($content);
        return self::html_handler("<meta charset='UTF-8'><pre>$content</pre>");
    }

    static function html_handler($content) {
        $dom = new DOMDocument();
        $dom->loadHTML(self::strip_invalid_xml_chars($content));
        return $dom->saveXML();
    }

    static function docx_handler($content, $extension = 'docx') {

        if (! self::executable_exists(self::$libreoffice)) {
            throw new Exception("libreoffice executable is not found");
        }

        # prepare file names.
        $workspace   = self::get_workspace();
        $origin_file = $workspace . DIRECTORY_SEPARATOR . "document.$extension";
        $out_dir     = $workspace . DIRECTORY_SEPARATOR . 'html';
        $out_file    = $out_dir . DIRECTORY_SEPARATOR . "document.html";

        file_put_contents($origin_file, $content);

        # execute the command
        $cmd = sprintf("export HOME={$workspace} && " . self::$libreoffice." --headless --convert-to html:HTML %s --outdir %s 2>&1",
            escapeshellarg($origin_file),
            escapeshellarg($out_dir)
        );
        exec($cmd, $output, $exit_code);

        # sometimes libreoffice returns 139 even if it already did conversion fine
        $has_error = ($exit_code !== 0) && ($exit_code !== 139);
        if ($has_error) {
            throw new ExecFailedException("Failed to convert to html", $exit_code, $cmd, $output);
        }

        $dom = new DOMDOcument();
        $dom->loadHTML(file_get_contents($out_file));
        $html = $dom->saveXML($dom->documentElement);
        return $html;
    }

    static function hwp_handler($content) {
        if (! self::executable_exists(self::$hwp5html)) {
            throw new Exception("hwp5html executable is not found.");
        }

        # prepare file names
        $workspace = self::get_workspace();
        $origin_file = $workspace . DIRECTORY_SEPARATOR . "document.hwp";
        $out_dir     = $workspace . DIRECTORY_SEPARATOR . "html";
        $out_file    = $out_dir . DIRECTORY_SEPARATOR . 'index.xhtml';

        file_put_contents($origin_file, $content);

        # execute the command
        $cmd = sprintf(self::$hwp5html." %s %s 2>&1",
            escapeshellarg($origin_file),
            $out_dir
        );
        exec($cmd, $output, $exit_code);
        $has_error = ($exit_code !== 0);
        if ($has_error) {
            throw new ExecFailedException("Failed to convert to html", $exit_code, $cmd, $output);
        }

        return file_get_contents($out_file);
    }

    static function doc_handler($content) {
        return self::docx_handler($content, 'doc');
    }

    static function strip_invalid_xml_chars($content) {
        // See http://www.w3.org/TR/xml/#charsets
        return preg_replace('/
            [^
                \x{9}
                \x{A}
                \x{D}
                \x{20}-\x{D7FF}
                \x{E000}-\x{FFFD}
                \x{10000}-\x{10FFF}
            ]/ux', '',
            $content
        );
    }

    private static function get_workspace() {

        // 랜덤 파일 생성해서 이름을 받고. 기존에 exec로 output을 받던 것은 맥에서 호환성 문제가 발생해 사용하지 않음.
        $workspace = tempnam(sys_get_temp_dir(), 'a2html.');

        // 파일은 지우고, 디렉토리를 만듦.
        unlink($workspace);
        if( ! is_dir($workspace)){
            mkdir($workspace, 0777, true);
        }
        register_shutdown_function(function () use ($workspace) {
            call_user_func(['Anything2html', 'rm_rf'], $workspace);
        });
        return $workspace;
    }

    private static function rm_rf($path) {
        if(strstr($path, sys_get_temp_dir())){
            @exec(sprintf("rm -rf %s", escapeshellarg($path)));
        }else{
            echo 'You do not rm folder that is not temp.';
            exit;
        }
    }

}

/**
 * Class ExecFailedException
 */
class ExecFailedException extends Exception {
    public $cmd;
    public $output;

    /**
     * @param string    $message
     * @param int       $code
     * @param string    $cmd
     * @param string    $output
     * @param Exception $previous
     */
    function __construct($message, $code, $cmd, $output, $previous = null) {
        $this->cmd    = $cmd;
        $this->output = implode("\n", $output);
        parent::__construct($message, $code, $previous);
    }
}